| 摘要: |
| 针对真实信道使无线通信信号发生相位偏移导致信号识别率下降的问题，提出了相位变换算法。该算法将信号输入神经网络，通过Flatten层和Dense层估计出相位参数，再利用参数变换器完成相位变换，减轻相位偏移对调制识别准确率的影响，提高了调制识别的准确率。同时提出一种卷积双向长短期记忆(Convolutional Neural Network-Bidirectional Long Short-Term Memory,CNN-BiLSTM)网络，其中CNN用于提取信号的高维特征，BiLSTM用于提取信号的双向时间特征。所提出的调制信号分类模型识别率达到96.7%，相较于卷积神经网络(Convolutional Neural Network,CNN)提高了12%。 |
| 关键词: 自动调制识别 相位变换 卷积双向长短期记忆网络 数据预处理 |
| DOI：10.20079/j.issn.1001-893x.230918001 |
|
| 基金项目:江苏省研究生创新项目(SJCX24_0445);无锡市社会发展科技示范工程项目(N20191008) |
|
| An Automatic Modulation Recognition Algorithm Based on Phase Transformation and CNN-BiLSTM |
| HU Guole,a,a,b |
| (1.Jiangsu Key Laboratory of Meteorological Observation and Information Processing,Nanjing University of Information Science & Technology,Nanjing 210044,China;2a.Jiangsu Industrial Environmental Hazard Monitoring and Evaluation Engineering Research Center;2b.School of Automation,Wuxi University,Wuxi 214105,China) |
| Abstract: |
| To solve the problem of the decrease of signal recognition rate in the real channel caused by the phase shift of wireless communication signal,a phase transformation algorithm is proposed.This algorithm inputs signals into a neural network,estimates phase parameters through the Flatten and Dense layers,and then uses a parameter transformer to complete phase transformation,thus reducing the impact of phase shift on modulation recognition accuracy and improving modulation recognition accuracy.Simultaneously,a convolutional neural network-bidirectional long short-term memory(CNN-BiLSTM) network is proposed,where CNN is used to extract high-dimensional features of signals and BiLSTM is used to extract bidirectional temporal features of signals.The recognition rate of the proposed modulated signal classification model reaches 96.7%,which is 12% higher than that of the CNN. |
| Key words: automatic modulation recognition phase transformation convolutional neural network-bidirectional long short-term memory network data preprocessing |
